In this paper, we propose a novel Reinforcement Learning (RL) framework for problems with continuous action spaces: Action Quantization from Demonstrations (AQuaDem). The proposed approach consists in learning a discretization of continuous action spaces from human demonstrations. This discretization returns a set of plausible actions (in light of the demonstrations) for each input state, thus capturing the priors of the demonstrator and their multimodal behavior. By discretizing the action space, any discrete action deep RL technique can be readily applied to the continuous control problem. Experiments show that the proposed approach outperforms state-of-the-art methods such as SAC in the RL setup, and GAIL in the Imitation Learning setup. We provide a website with interactive videos: https://google-research.github.io/aquadem/ and make the code available: https://github.com/google-research/google-research/tree/master/aquadem.
@article{arxiv.2110.10149,
title = {Continuous Control with Action Quantization from Demonstrations},
author = {Robert Dadashi and Léonard Hussenot and Damien Vincent and Sertan Girgin and Anton Raichuk and Matthieu Geist and Olivier Pietquin},
journal= {arXiv preprint arXiv:2110.10149},
year = {2022}
}